With digital education platforms generating data on how millions of students are learning, they are also creating a veritable gold mine of information for researchers looking to improve education.
Ethical and legal issues stand in the way. In other words, how to share that data responsibly without leaving students open to the possibility of their personal information being exposed to outside parties.
Now a consortium of education researchers and learning platforms is developing the desired solution. Researchers never see the actual data.
A project called SafeInsights, led by Rice University's OpenStax, is funded by a $90 million grant from the National Science Foundation over five years.
SafeInsights' goal is to serve as a bridge between learning platforms and research partners, and work with collaborators to help shape how the exchange works to protect student privacy.
“Under normal circumstances, you end up taking data from learning websites and apps and making it available to researchers to study, analyze, and learn from,” says JP Slavinsky, Managing Director of SafeInsights and Technical Director of OpenStax. “Instead, we are putting researchers’ questions into the data. “Because the data stays where it already is, it creates a safe research environment that makes it easier for schools and platforms to participate.”
Deeper insights at scale
Another way to think of SafeInsights as a telescope, says Slavinsky and his colleague Richard Baraniuk, founder and director of OpenStax, which publishes open access course materials. This allows researchers to look at vast amounts of data from learning platforms such as the University of Pennsylvania's Massive Online Open Courses and Quill.org.
Researchers develop questions and then translate those questions into computer code that can sift through the data and feed them into a learning platform. Once results are generated, they are returned to researchers without the need to share the data directly.
“This is really a partnership where researchers come together with schools and platforms and collectively work to solve some problems of interest,” Slavinsky says. “We are making the telescope available so that others can bring their research agendas and the questions they want to answer. So we're less concerned with specifically what questions to ask and more focused on making sure we can answer as many questions as possible.”
One of the reasons this model is so powerful is because of how it can increase the scale at which educational research is conducted, Baraniuk says. He explains that there are many studies with small sample sizes of about 50 college students in psychology classes.
“Most studies are on first-year college students. Yes? Well, it’s not representative of the huge breadth of diverse students,” Baraniuk says. “The only way to see that breadth is to conduct large-scale research, so the first pillar of SafeInsights is partnering with digital education websites and apps that host literally millions of students every day.”
Another aspect in which the project opens new doors for researchers is the diversity of the student population represented by the learning platform partners, which include educational apps for reading, writing, and science along with learning management systems.
“The idea is that if we put all these puzzle pieces together, we can get a more complete picture of these students on a very large scale,” Baraniuk says. “Our biggest goal is to try to remove as much friction as possible so that more useful research can happen. More research-supported teaching methods and educational technologies can then be applied in practice. But how can we remove that friction and actually keep everything safe?”
Build trust, protect privacy
Before the study gets underway, SafeInsights partners at the Future of Privacy Forum are helping develop policies that will shape how the program protects students' data.
The goal is to embed privacy features into how everything works, says John Verdi, senior vice president of policy at the Future of Privacy Forum. Part of that is helping develop what he calls the “data plane,” a process that allows researchers to query data in learning platforms without having to access it directly. Other aspects include helping develop a review process for how research projects are selected, training researchers on privacy, and publishing lessons learned about how to operate with privacy as a top priority.
Regarding the training aspect, he said, “Even if you have good technical safeguards in place and you do good ethical review, ultimately researchers make their own decisions about how to use the system responsibly. They need to understand how the system works.”
In education, student data privacy is generally “woefully underfunded,” he says, but protecting their information helps students trust their learning platforms and ultimately creates research opportunities like SafeInsights.
“Putting data protection in the hands of students and parents is the wrong responsibility,” says Verdi. “What we need to do instead is build a digital infrastructure that fundamentally respects privacy. [that] It provides assurance that information is kept confidential and used ethically.”